Spaces:
Running
on
A100
Running
on
A100
Update app.py
Browse files
app.py
CHANGED
@@ -4,22 +4,21 @@ import gradio as gr
|
|
4 |
from PIL import Image
|
5 |
import torch
|
6 |
import spaces
|
|
|
7 |
|
8 |
-
#
|
9 |
-
# Environment
|
10 |
-
# --------------------------
|
11 |
MODEL_ID = os.environ.get("MODEL_ID", "inference-net/ClipTagger-12b")
|
12 |
-
HF_TOKEN = os.environ.get("HF_TOKEN")
|
13 |
-
|
14 |
-
DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
|
15 |
-
DTYPE = torch.bfloat16 if torch.cuda.is_available() else torch.float32
|
16 |
-
|
17 |
TEMP = 0.1
|
18 |
MAX_NEW_TOKENS = 2000
|
19 |
|
20 |
-
#
|
21 |
-
|
22 |
-
|
|
|
|
|
|
|
|
|
23 |
SYSTEM_PROMPT = (
|
24 |
"You are an image annotation API trained to analyze YouTube video keyframes. "
|
25 |
"You will be given instructions on the output format, what to caption, and how to perform your job. "
|
@@ -56,77 +55,105 @@ Rules:
|
|
56 |
- Output **only the JSON**, no extra text or explanation.
|
57 |
"""
|
58 |
|
59 |
-
#
|
60 |
-
|
61 |
-
|
62 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
63 |
|
64 |
-
|
65 |
-
|
|
|
|
|
|
|
|
|
66 |
|
67 |
-
|
68 |
-
|
69 |
-
|
70 |
-
|
71 |
-
|
72 |
-
|
73 |
-
|
|
|
|
|
|
|
74 |
|
75 |
-
# Processor (has vision + tokenizer routing)
|
76 |
try:
|
77 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
78 |
MODEL_ID, token=HF_TOKEN, trust_remote_code=True, use_fast=True
|
79 |
)
|
80 |
-
|
81 |
-
|
82 |
-
|
|
|
|
|
|
|
83 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
84 |
|
85 |
-
# Model
|
86 |
-
model = AutoModelForCausalLM.from_pretrained(
|
87 |
-
MODEL_ID,
|
88 |
-
token=HF_TOKEN,
|
89 |
-
device_map="auto",
|
90 |
-
torch_dtype=DTYPE,
|
91 |
-
trust_remote_code=True,
|
92 |
-
)
|
93 |
-
|
94 |
-
# Tokenizer (fall back in case processor doesn't expose it)
|
95 |
-
tokenizer = getattr(processor, "tokenizer", None) or AutoTokenizer.from_pretrained(
|
96 |
-
MODEL_ID, token=HF_TOKEN, trust_remote_code=True, use_fast=True
|
97 |
-
)
|
98 |
-
|
99 |
-
except Exception as e:
|
100 |
-
LOAD_ERROR = f"{e}\n\n{traceback.format_exc()}"
|
101 |
-
|
102 |
-
# --------------------------
|
103 |
-
# Inference
|
104 |
-
# --------------------------
|
105 |
-
def _build_messages(image: Image.Image):
|
106 |
-
return [
|
107 |
-
{"role": "system", "content": [{"type": "text", "text": SYSTEM_PROMPT}]},
|
108 |
-
{"role": "user", "content": [{"type": "image", "image": image},
|
109 |
-
{"type": "text", "text": USER_PROMPT}]}
|
110 |
-
]
|
111 |
-
|
112 |
-
def _run(image: Image.Image) -> Tuple[str, Dict[str, Any], bool]:
|
113 |
if image is None:
|
114 |
return "Please upload an image.", None, False
|
115 |
-
if model is None or processor is None:
|
116 |
-
msg = (
|
117 |
-
"❌ Model failed to load.\n\n"
|
118 |
-
f"{LOAD_ERROR or 'Unknown error.'}\n"
|
119 |
-
"Check: MODEL_ID, HF_TOKEN, and that the repo includes processor + model shards."
|
120 |
-
)
|
121 |
-
return msg, None, False
|
122 |
|
123 |
-
#
|
124 |
-
if hasattr(
|
125 |
-
prompt =
|
126 |
-
_build_messages(image), add_generation_prompt=True, tokenize=False
|
127 |
-
)
|
128 |
else:
|
129 |
-
# Conservative fallback
|
130 |
msgs = _build_messages(image)
|
131 |
prompt = ""
|
132 |
for m in msgs:
|
@@ -137,93 +164,49 @@ def _run(image: Image.Image) -> Tuple[str, Dict[str, Any], bool]:
|
|
137 |
elif chunk["type"] == "image":
|
138 |
prompt += f"{role}: [IMAGE]\n"
|
139 |
|
140 |
-
|
141 |
-
inputs = processor(text=prompt, images=image, return_tensors="pt").to(model.device)
|
142 |
|
143 |
-
# Generation args
|
144 |
gen_kwargs = dict(
|
145 |
temperature=TEMP,
|
146 |
max_new_tokens=MAX_NEW_TOKENS,
|
147 |
)
|
148 |
-
#
|
149 |
-
|
150 |
-
|
151 |
-
|
152 |
-
from transformers.utils import is_torch_available
|
153 |
-
cfg_eos = getattr(model.config, "eos_token_id", None)
|
154 |
-
if isinstance(cfg_eos, (list, tuple)):
|
155 |
-
gen_kwargs["eos_token_id"] = list(cfg_eos)
|
156 |
-
elif eos_id is not None:
|
157 |
-
gen_kwargs["eos_token_id"] = eos_id
|
158 |
-
except Exception:
|
159 |
-
if eos_id is not None:
|
160 |
-
gen_kwargs["eos_token_id"] = eos_id
|
161 |
|
162 |
-
#
|
163 |
try:
|
164 |
gen_kwargs["response_format"] = {"type": "json_object"}
|
165 |
except Exception:
|
166 |
pass
|
167 |
|
168 |
with torch.inference_mode():
|
169 |
-
|
170 |
|
171 |
-
|
172 |
-
|
173 |
-
|
174 |
-
else:
|
175 |
-
text = tokenizer.decode(out_ids[0], skip_special_tokens=True)
|
176 |
|
177 |
-
# Trim any echoed prompt
|
178 |
if USER_PROMPT in text:
|
179 |
text = text.split(USER_PROMPT)[-1].strip()
|
180 |
|
181 |
-
|
182 |
-
|
183 |
-
parsed = json.loads(text)
|
184 |
return json.dumps(parsed, indent=2), parsed, True
|
185 |
-
|
186 |
-
m = re.search(r"\{(?:[^{}]|(?R))*\}", text, flags=re.DOTALL)
|
187 |
-
if m:
|
188 |
-
try:
|
189 |
-
parsed = json.loads(m.group(0))
|
190 |
-
return json.dumps(parsed, indent=2), parsed, True
|
191 |
-
except Exception:
|
192 |
-
pass
|
193 |
-
# Return raw text to help debug prompt adherence if needed
|
194 |
-
return text, None, False
|
195 |
-
|
196 |
-
# --------------------------
|
197 |
-
# Spaces GPU entry + warmup
|
198 |
-
# --------------------------
|
199 |
-
@spaces.GPU
|
200 |
-
def annotate_image(pil: Image.Image):
|
201 |
-
return _run(pil)
|
202 |
|
|
|
203 |
@spaces.GPU(duration=60)
|
204 |
def _warmup():
|
205 |
-
if model is None or processor is None:
|
206 |
-
return "skip"
|
207 |
try:
|
208 |
-
|
209 |
-
_ = _run(dummy)
|
210 |
-
return "ok"
|
211 |
except Exception as e:
|
212 |
return f"warmup error: {e}"
|
213 |
|
214 |
-
|
215 |
-
|
216 |
-
|
217 |
-
pass
|
218 |
-
|
219 |
-
# --------------------------
|
220 |
-
# UI
|
221 |
-
# --------------------------
|
222 |
-
with gr.Blocks(theme=gr.themes.Soft(), analytics_enabled=False, title="Keyframe Annotator (Gemma-3 VLM)") as demo:
|
223 |
-
gr.Markdown("# Keyframe Annotator (Gemma-3-12B FT)\nUpload an image to get **strict JSON** annotations.")
|
224 |
-
if LOAD_ERROR:
|
225 |
-
with gr.Accordion("Startup Error Details", open=False):
|
226 |
-
gr.Markdown(f"```\n{LOAD_ERROR}\n```")
|
227 |
|
228 |
with gr.Row():
|
229 |
with gr.Column(scale=1):
|
@@ -234,10 +217,12 @@ with gr.Blocks(theme=gr.themes.Soft(), analytics_enabled=False, title="Keyframe
|
|
234 |
out_json = gr.JSON(label="Parsed JSON")
|
235 |
ok_flag = gr.Checkbox(label="Valid JSON", value=False, interactive=False)
|
236 |
|
237 |
-
def on_click(img):
|
238 |
-
text, js, ok = _run(img)
|
239 |
-
return text, js, ok
|
240 |
-
|
241 |
btn.click(annotate_image, inputs=[image], outputs=[out_text, out_json, ok_flag])
|
242 |
|
|
|
|
|
|
|
|
|
|
|
|
|
243 |
demo.queue(max_size=32).launch()
|
|
|
4 |
from PIL import Image
|
5 |
import torch
|
6 |
import spaces
|
7 |
+
from transformers import AutoProcessor, AutoTokenizer, AutoModelForCausalLM, AutoConfig
|
8 |
|
9 |
+
# --------- ENV / PARAMS ----------
|
|
|
|
|
10 |
MODEL_ID = os.environ.get("MODEL_ID", "inference-net/ClipTagger-12b")
|
11 |
+
HF_TOKEN = os.environ.get("HF_TOKEN") # put this in Space -> Settings -> Variables & secrets
|
|
|
|
|
|
|
|
|
12 |
TEMP = 0.1
|
13 |
MAX_NEW_TOKENS = 2000
|
14 |
|
15 |
+
# Lazy globals (ZeroGPU-safe)
|
16 |
+
_processor: Any = None
|
17 |
+
_tokenizer: Any = None
|
18 |
+
_model: Any = None
|
19 |
+
_last_load_error: str | None = None
|
20 |
+
|
21 |
+
# --------- PROMPTS (yours) ----------
|
22 |
SYSTEM_PROMPT = (
|
23 |
"You are an image annotation API trained to analyze YouTube video keyframes. "
|
24 |
"You will be given instructions on the output format, what to caption, and how to perform your job. "
|
|
|
55 |
- Output **only the JSON**, no extra text or explanation.
|
56 |
"""
|
57 |
|
58 |
+
# --------- HELPERS ----------
|
59 |
+
def _json_extract(text: str):
|
60 |
+
"""Strict parse -> top-level {...} fallback."""
|
61 |
+
try:
|
62 |
+
return json.loads(text)
|
63 |
+
except Exception:
|
64 |
+
m = re.search(r"\{(?:[^{}]|(?R))*\}", text, flags=re.DOTALL)
|
65 |
+
if m:
|
66 |
+
try:
|
67 |
+
return json.loads(m.group(0))
|
68 |
+
except Exception:
|
69 |
+
pass
|
70 |
+
return None
|
71 |
|
72 |
+
def _build_messages(image: Image.Image):
|
73 |
+
return [
|
74 |
+
{"role": "system", "content": [{"type": "text", "text": SYSTEM_PROMPT}]},
|
75 |
+
{"role": "user", "content": [{"type": "image", "image": image},
|
76 |
+
{"type": "text", "text": USER_PROMPT}]}
|
77 |
+
]
|
78 |
|
79 |
+
# --------- ZERO-GPU LAZY LOADER ----------
|
80 |
+
@spaces.GPU
|
81 |
+
def _ensure_loaded() -> str:
|
82 |
+
"""
|
83 |
+
Load the model only when a ZeroGPU worker with a GPU is attached.
|
84 |
+
Tries quantized path first (compressed-tensors), then falls back to unquantized.
|
85 |
+
"""
|
86 |
+
global _processor, _tokenizer, _model, _last_load_error
|
87 |
+
if _model is not None and _processor is not None:
|
88 |
+
return "already_loaded"
|
89 |
|
|
|
90 |
try:
|
91 |
+
# Sanity: config should be gemma3 causal VLM (not CLIP)
|
92 |
+
cfg = AutoConfig.from_pretrained(MODEL_ID, token=HF_TOKEN, trust_remote_code=True)
|
93 |
+
if "clip" in cfg.__class__.__name__.lower():
|
94 |
+
raise RuntimeError(
|
95 |
+
f"MODEL_ID '{MODEL_ID}' resolves to CLIP/encoder config; need a causal VLM checkpoint."
|
96 |
+
)
|
97 |
+
|
98 |
+
# Try quantized (as per your config)
|
99 |
+
_processor = AutoProcessor.from_pretrained(
|
100 |
MODEL_ID, token=HF_TOKEN, trust_remote_code=True, use_fast=True
|
101 |
)
|
102 |
+
_model = AutoModelForCausalLM.from_pretrained(
|
103 |
+
MODEL_ID,
|
104 |
+
token=HF_TOKEN,
|
105 |
+
device_map="auto",
|
106 |
+
torch_dtype=torch.bfloat16 if torch.cuda.is_available() else torch.float32,
|
107 |
+
trust_remote_code=True,
|
108 |
)
|
109 |
+
_tokenizer = getattr(_processor, "tokenizer", None) or AutoTokenizer.from_pretrained(
|
110 |
+
MODEL_ID, token=HF_TOKEN, trust_remote_code=True, use_fast=True
|
111 |
+
)
|
112 |
+
_last_load_error = None
|
113 |
+
return "ok_quant"
|
114 |
+
except Exception as e:
|
115 |
+
# Fallback: disable quantization (more VRAM)
|
116 |
+
if "compressed_tensors" in str(e):
|
117 |
+
try:
|
118 |
+
_processor = AutoProcessor.from_pretrained(
|
119 |
+
MODEL_ID, token=HF_TOKEN, trust_remote_code=True, use_fast=True
|
120 |
+
)
|
121 |
+
_model = AutoModelForCausalLM.from_pretrained(
|
122 |
+
MODEL_ID,
|
123 |
+
token=HF_TOKEN,
|
124 |
+
device_map="auto",
|
125 |
+
torch_dtype=torch.bfloat16 if torch.cuda.is_available() else torch.float32,
|
126 |
+
trust_remote_code=True,
|
127 |
+
quantization_config=None, # force dequantized load
|
128 |
+
)
|
129 |
+
_tokenizer = getattr(_processor, "tokenizer", None) or AutoTokenizer.from_pretrained(
|
130 |
+
MODEL_ID, token=HF_TOKEN, trust_remote_code=True, use_fast=True
|
131 |
+
)
|
132 |
+
_last_load_error = None
|
133 |
+
return "ok_dequant"
|
134 |
+
except Exception as e2:
|
135 |
+
_last_load_error = f"{e}\n\nFallback failed:\n{e2}\n{traceback.format_exc()}"
|
136 |
+
_processor = _tokenizer = _model = None
|
137 |
+
return "fail"
|
138 |
+
else:
|
139 |
+
_last_load_error = f"{e}\n{traceback.format_exc()}"
|
140 |
+
_processor = _tokenizer = _model = None
|
141 |
+
return "fail"
|
142 |
+
|
143 |
+
# --------- INFERENCE ----------
|
144 |
+
@spaces.GPU
|
145 |
+
def annotate_image(image: Image.Image) -> Tuple[str, Dict[str, Any] | None, bool]:
|
146 |
+
status = _ensure_loaded()
|
147 |
+
if status == "fail":
|
148 |
+
return f"❌ Load error:\n{_last_load_error}", None, False
|
149 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
150 |
if image is None:
|
151 |
return "Please upload an image.", None, False
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
152 |
|
153 |
+
# Prompt assembly
|
154 |
+
if hasattr(_processor, "apply_chat_template"):
|
155 |
+
prompt = _processor.apply_chat_template(_build_messages(image), add_generation_prompt=True, tokenize=False)
|
|
|
|
|
156 |
else:
|
|
|
157 |
msgs = _build_messages(image)
|
158 |
prompt = ""
|
159 |
for m in msgs:
|
|
|
164 |
elif chunk["type"] == "image":
|
165 |
prompt += f"{role}: [IMAGE]\n"
|
166 |
|
167 |
+
inputs = _processor(text=prompt, images=image, return_tensors="pt").to(_model.device)
|
|
|
168 |
|
|
|
169 |
gen_kwargs = dict(
|
170 |
temperature=TEMP,
|
171 |
max_new_tokens=MAX_NEW_TOKENS,
|
172 |
)
|
173 |
+
# respect multiple eos ids if present
|
174 |
+
eos = getattr(_model.config, "eos_token_id", None)
|
175 |
+
if eos is not None:
|
176 |
+
gen_kwargs["eos_token_id"] = eos
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
177 |
|
178 |
+
# Try JSON-only output (if supported)
|
179 |
try:
|
180 |
gen_kwargs["response_format"] = {"type": "json_object"}
|
181 |
except Exception:
|
182 |
pass
|
183 |
|
184 |
with torch.inference_mode():
|
185 |
+
out = _model.generate(**inputs, **gen_kwargs)
|
186 |
|
187 |
+
text = (_processor.decode(out[0], skip_special_tokens=True)
|
188 |
+
if hasattr(_processor, "decode")
|
189 |
+
else _tokenizer.decode(out[0], skip_special_tokens=True))
|
|
|
|
|
190 |
|
|
|
191 |
if USER_PROMPT in text:
|
192 |
text = text.split(USER_PROMPT)[-1].strip()
|
193 |
|
194 |
+
parsed = _json_extract(text)
|
195 |
+
if isinstance(parsed, dict):
|
|
|
196 |
return json.dumps(parsed, indent=2), parsed, True
|
197 |
+
return text, None, False
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
198 |
|
199 |
+
# Optional: quick warmup to validate loading on first worker
|
200 |
@spaces.GPU(duration=60)
|
201 |
def _warmup():
|
|
|
|
|
202 |
try:
|
203 |
+
return _ensure_loaded()
|
|
|
|
|
204 |
except Exception as e:
|
205 |
return f"warmup error: {e}"
|
206 |
|
207 |
+
# --------- UI ----------
|
208 |
+
with gr.Blocks(theme=gr.themes.Soft(), analytics_enabled=False, title="Keyframe Annotator (ZeroGPU)") as demo:
|
209 |
+
gr.Markdown("# Keyframe Annotator (Gemma-3-12B FT · ZeroGPU)\nUpload an image to get **strict JSON** annotations.")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
210 |
|
211 |
with gr.Row():
|
212 |
with gr.Column(scale=1):
|
|
|
217 |
out_json = gr.JSON(label="Parsed JSON")
|
218 |
ok_flag = gr.Checkbox(label="Valid JSON", value=False, interactive=False)
|
219 |
|
|
|
|
|
|
|
|
|
220 |
btn.click(annotate_image, inputs=[image], outputs=[out_text, out_json, ok_flag])
|
221 |
|
222 |
+
# fire a non-blocking warmup
|
223 |
+
try:
|
224 |
+
_ = _warmup()
|
225 |
+
except Exception:
|
226 |
+
pass
|
227 |
+
|
228 |
demo.queue(max_size=32).launch()
|